Statistical Methods for the Analysis of ChlP-chip Data

描述（由申请人提供）：随着许多基因组测序项目即将结束，剩下的最大挑战是理解这些序列中编码的信息。识别转录因子（TF）和它们的DMA结合位点之间的相互作用是这一挑战的一个组成部分。这些相互作用控制着细胞功能的关键步骤，它们的功能障碍可以显著促进各种疾病的进展。ChIP芯片实验耦合染色质免疫沉淀与DMA微阵列分析已成为强大的工具，全基因组的识别和转录因子结合位点的表征。这些实验产生大量的噪声数据，重复次数很少，因此需要创新的强大的统计分析方法。该提案的目标是开发、评估和传播用于分析ChIP芯片实验数据的统计方法。这些目标将通过以下四个具体目标来实现：（1）开发用于检测TF结合区域的鲁棒概率方法。这些方法将利用平铺阵列上探针之间的共同信息，以增加小样本量的功效。(2)扩展Aim-1中的方法，以处理探针序列重叠且附近探针的观察结果显示出长程空间依赖性的阵列设计。因此，我们将制定严格的统计推断程序一般平铺阵列设计。(3)开发一个自适应框架，将ChIP芯片实验的定量信息纳入基序发现。这将连接ChIP-芯片数据分析的第一阶段，即结合区域的鉴定，与下游序列分析，从而提高基序发现任务的灵敏度和特异性。(4)在统计软件包中实施作为本研究一部分开发的统计方法。由此产生的软件包将以独立版本和作为Bioconductor项目的一部分提供给科学界，Bioconductor项目是一个用于分析基因组数据的开源和开发软件项目。成功完成拟议的研究将导致大幅改善ChIP芯片实验分析的统计方法。

DESCRIPTION (provided by applicant): With many genome-sequencing projects coming to an end, the biggest remaining challenge is to comprehend the information encoded in these sequences. Identifying interactions between transcription factors (TFs) and their DMA binding sites is an integral part of this challenge. These interactions control critical steps in cell functions, and their dysfunction can significantly contribute to the progression of various diseases. ChlP-chip experiments that couple chromatin immunoprecipitation with DMA microarray analysis have become powerful tools for the genome-wide identification and characterization of transcription factor binding sites. These experiments produce massive amounts of noisy data with small number of replicates and therefore require innovative robust statistical analysis methods. The objectives of this proposal are to develop, evaluate and disseminate statistical methods for analyzing data from ChlP-chip experiments. These objectives will be accomplished through four specific aims: (1) Development of robust probabilistic methods for detecting TF bound regions. These methods will utilize the information common across probes on tiling arrays to increase power in small sample sizes. (2) Extension of the methods in Aim-1 to deal with array designs where probe sequences overlap and observations from nearby probes exhibit long-range spatial dependencies. As a result, we will develop rigorous statistical inference procedures for general tiling array designs. (3) Development of an adaptive framework for incorporating quantitative information from ChlP-chip experiments into motif finding. This will connect the first stage of the ChlP-chip data analysis, namely identification of the bound regions, with the downstream sequence analysis thereby boosting the sensitivity and specificity of the motif finding task. (4) Implementation of the statistical methods developed as part of this research in statistical packages. The resulting packages will be available to the scientific community both in stand-alone versions and as part of the Bioconductor Project which is an open source and development software project for the analysis of the genomic data. Successful completion of the proposed research will result in substantially improved statistical methods for the analysis of ChlP-chip experiments.